Hi Mathias,
On 9/11/2023 5:15 PM, Wesley Cheng wrote:
Hi Mathias,
On 9/11/2023 6:50 AM, Mathias Nyman wrote:
On 1.9.2023 3.15, Wesley Cheng wrote:
There is a 120ms delay implemented for allowing the XHCI host
controller to
detect a U3 wakeup pulse. The intention is to wait for the device to
retry
the wakeup event if the USB3 PORTSC doesn't reflect the RESUME link
status
by the time it is checked. As per the USB3 specification:
tU3WakeupRetryDelay ("Table 7-12. LTSSM State Transition Timeouts")
This would allow the XHCI resume sequence to determine if the root hub
needs to be also resumed. However, in case there is no device
connected,
or if there is only a HSUSB device connected, this delay would still
affect
the overall resume timing.
Since this delay is solely for detecting U3 wake events (USB3 specific)
then ignore this delay for the disconnected case and the HSUSB connected
only case.
Thanks, makes sense
Signed-off-by: Wesley Cheng <quic_wcheng@xxxxxxxxxxx>
---
drivers/usb/host/xhci.c | 26 +++++++++++++++++++++++++-
1 file changed, 25 insertions(+), 1 deletion(-)
diff --git a/drivers/usb/host/xhci.c b/drivers/usb/host/xhci.c
index e1b1b64a0723..640db6a4e686 100644
--- a/drivers/usb/host/xhci.c
+++ b/drivers/usb/host/xhci.c
@@ -805,6 +805,24 @@ static void xhci_disable_hub_port_wake(struct
xhci_hcd *xhci,
spin_unlock_irqrestore(&xhci->lock, flags);
}
+static bool xhci_usb3_dev_connected(struct xhci_hcd *xhci)
+{
+ struct xhci_port **ports;
+ int port_index;
+ u32 portsc;
+
+ port_index = xhci->usb3_rhub.num_ports;
+ ports = xhci->usb3_rhub.ports;
+
+ while (port_index--) {
+ portsc = readl(ports[port_index]->addr);
+ if (portsc & (portsc & PORT_CONNECT))
+ return true;
+ }
+
+ return false;
+}
+
This is one way, but we can probably avoid re-reading all the usb3
portsc registers
by checking if any bit is set in either:
// bitfield, set if xhci usb3 port neatly set to U3 with a hub request
xhci->usb3_rhub.bus_state.suspended_ports
// bitfield, set if xhci usb3 port is forced to U3 during xhci suspend.
xhci->usb3_rhub.bus_state.bus_suspended
But haven't checked this works in all corner cases.
Thanks for the suggestion. I think I also looked at seeing if we could
use the suspended_ports param, and it was missing one of the use cases
we had. I haven't thought on combining it with the bus_suspend param
also to see if it could work. Let me give it a try, and I'll get back
to you.
So in one of our normal use cases, which is to use an USB OTG adapter
with our devices, we can have this connected with no device. In this
situation, the XHCI HCD and root hub are enumerated, and is in a state
where nothing is connected to the port. I added a print to the
xhci_resume() path to check the status of "suspended_ports" and
"bus_suspended" and they seem to reflect the same status as when there
is something connected (to a device that supports autosuspend). Here's
some pointers I've found on why these parameters may not work:
1. bus_suspended is only set (for the bus) if we reach the
bus_suspend() callback from USB HCD if the link is still in U0. If USB
autosuspend is enabled, then the suspending of the root hub udev, would
have caused a call to suspend the port (usb_port_suspend()), and that
would set "suspended_ports" and placed the link in U3 already.
2. "suspended_ports" can't differentiate if a device is connected or not
after plugging in a USB3 device that has autosuspend enabled. It looks
like on device disconnection, the suspended_ports isn't cleared for that
port number. It is only cleared during the resume path where a get port
status is queried:
static void xhci_get_usb3_port_status(struct xhci_port *port, u32 *status,
u32 portsc)
{
...
/* USB3 specific wPortStatus bits */
if (portsc & PORT_POWER) {
*status |= USB_SS_PORT_STAT_POWER;
/* link state handling */
if (link_state == XDEV_U0)
bus_state->suspended_ports &= ~(1 << portnum);
}
IMO, this seems kind of weird, because the PLS shows that the port is in
the RxDetect state, so it technically isn't suspended. If you think we
should clear suspended_ports on disconnect, then I think we can also
change the logic to rely on it for avoiding the unnecessary delay in
xhci_resume().
Thanks
Wesley Cheng
